sync_data.sh

#!/bin/sh

# 设置默认值
: ${APP_HOME:=/OpenList}
: ${STORAGE_PATH:=$APP_HOME/data}
: ${SYNC_INTERVAL:=7200}
: ${MAX_BACKUPS:=50}

# 检查必要环境变量
if [ -z "$HF_TOKEN" ] || [ -z "$DATASET_ID" ]; then
    echo "警告: 未检测到 HF_TOKEN 或 DATASET_ID，备份功能不可用"
    if [ "$REQUIRE_BACKUP" = "true" ]; then
        echo "错误: 备份为必需功能，但缺少必要环境变量"
        exit 1
    fi
    # 在没有备份功能的情况下直接运行服务
    exec ./openlist server
fi

# 激活虚拟环境
source /opt/venv/bin/activate

# 生成同步脚本
cat > hf_sync.py << 'EOL'
from huggingface_hub import HfApi
import sys
import os
import tarfile
import tempfile
import time
from datetime import datetime

def manage_backups(api, repo_id, max_files):
    try:
        files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
        backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
        backup_files.sort()
        
        if len(backup_files) >= max_files:
            files_to_delete = backup_files[:len(backup_files) - max_files + 1]
            for file_to_delete in files_to_delete:
                try:
                    api.delete_file(
                        path_in_repo=file_to_delete,
                        repo_id=repo_id,
                        repo_type="dataset"
                    )
                    print(f'已删除旧备份: {file_to_delete}')
                except Exception as e:
                    print(f'删除 {file_to_delete} 时出错: {str(e)}')
    except Exception as e:
        print(f'管理备份时出错: {str(e)}')

def upload_backup(file_path, file_name, token, repo_id, max_files):
    api = HfApi(token=token)
    try:
        print(f"正在上传备份 {file_name}...")
        start_time = time.time()
        
        api.upload_file(
            path_or_fileobj=file_path,
            path_in_repo=file_name,
            repo_id=repo_id,
            repo_type="dataset"
        )
        
        upload_time = time.time() - start_time
        print(f"成功上传 {file_name} (耗时 {upload_time:.2f}秒)")
        manage_backups(api, repo_id, max_files)
        return True
    except Exception as e:
        print(f"上传文件出错: {str(e)}")
        return False

def download_latest_backup(token, repo_id, extract_path, max_retries=3):
    for attempt in range(max_retries):
        try:
            api = HfApi(token=token)
            files = api.list_repo_files(repo_id=repo_id, repo_type="dataset")
            backup_files = [f for f in files if f.startswith('backup_') and f.endswith('.tar.gz')]
            
            if not backup_files:
                print("未找到任何备份文件")
                return False
                
            latest_backup = sorted(backup_files)[-1]
            print(f"正在下载最新备份: {latest_backup}")
            
            with tempfile.TemporaryDirectory() as temp_dir:
                filepath = api.hf_hub_download(
                    repo_id=repo_id,
                    filename=latest_backup,
                    repo_type="dataset",
                    local_dir=temp_dir
                )
                
                if filepath and os.path.exists(filepath):
                    print("正在解压备份文件...")
                    with tarfile.open(filepath, 'r:gz') as tar:
                        tar.extractall(extract_path)
                    print(f"已成功恢复备份: {latest_backup}")
                    return True
        except Exception as e:
            print(f"下载备份出错 (尝试 {attempt + 1}/${max_retries}): {str(e)}")
            if attempt < max_retries - 1:
                wait_time = (attempt + 1) * 10
                print(f"等待 {wait_time}秒后重试...")
                time.sleep(wait_time)
    
    return False

def super_squash_history(token, repo_id):
    try:
        api = HfApi(token=token)
        print("正在合并历史提交...")
        api.super_squash_history(repo_id=repo_id, repo_type="dataset")
        print("历史合并完成。")
    except Exception as e:
        print(f"合并历史出错: {str(e)}")

if __name__ == "__main__":
    action = sys.argv[1]
    token = sys.argv[2]
    repo_id = sys.argv[3]
    
    if action == "upload":
        file_path = sys.argv[4]
        file_name = sys.argv[5]
        max_files = int(sys.argv[6]) if len(sys.argv) > 6 else 50
        upload_backup(file_path, file_name, token, repo_id, max_files)
    elif action == "download":
        extract_path = sys.argv[4] if len(sys.argv) > 4 else '.'
        download_latest_backup(token, repo_id, extract_path)
    elif action == "super_squash":
        super_squash_history(token, repo_id)
EOL

# 首次启动时从 HuggingFace 下载最新备份
if [ "$SKIP_INITIAL_DOWNLOAD" != "true" ]; then
    echo "正在尝试从 HuggingFace 下载最新备份..."
    if ! python hf_sync.py download "${HF_TOKEN}" "${DATASET_ID}" "${STORAGE_PATH}"; then
        echo "警告: 初始备份下载失败，继续使用空数据目录"
        mkdir -p "${STORAGE_PATH}"
    fi
fi

# 同步函数
sync_data() {
    while true; do
        echo "[$(date '+%Y-%m-%d %H:%M:%S')] 开始同步流程"
        
        # 检查数据目录
        if [ ! -d "${STORAGE_PATH}" ]; then
            echo "错误: 存储目录 ${STORAGE_PATH} 不存在"
            sleep 60
            continue
        fi
        
        # 创建备份
        timestamp=$(date +%Y%m%d_%H%M%S)
        backup_file="backup_${timestamp}.tar.gz"
        temp_backup="/tmp/${backup_file}"
        
        echo "正在创建备份: ${backup_file}"
        start_time=$(date +%s)
        
        # 压缩目录
        if ! tar -czf "${temp_backup}" -C "$(dirname "${STORAGE_PATH}")" "$(basename "${STORAGE_PATH}")"; then
            echo "错误: 创建备份文件失败"
            rm -f "${temp_backup}"
            sleep ${SYNC_INTERVAL}
            continue
        fi
        
        # 检查备份文件
        if [ ! -f "${temp_backup}" ]; then
            echo "错误: 备份文件未创建成功"
            sleep ${SYNC_INTERVAL}
            continue
        fi
        
        # 上传备份
        echo "正在上传备份到 HuggingFace..."
        if python hf_sync.py upload "${HF_TOKEN}" "${DATASET_ID}" "${temp_backup}" "${backup_file}" "${MAX_BACKUPS}"; then
            echo "备份上传成功"
        else
            echo "警告: 备份上传失败"
        fi
        
        # 清理临时文件
        rm -f "${temp_backup}"
        
        # 每周合并历史
        SQUASH_FLAG_FILE="/tmp/last_squash_time"
        NOW=$(date +%s)
        SEVEN_DAYS=$((7*24*60*60))
        
        if [ ! -f "$SQUASH_FLAG_FILE" ]; then
            echo $NOW > "$SQUASH_FLAG_FILE"
            echo "首次合并历史提交..."
            python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
        else
            LAST=$(cat "$SQUASH_FLAG_FILE")
            DIFF=$((NOW - LAST))
            
            if [ $DIFF -ge $SEVEN_DAYS ]; then
                echo $NOW > "$SQUASH_FLAG_FILE"
                echo "距离上次合并已超过7天，正在合并历史提交..."
                python hf_sync.py super_squash "${HF_TOKEN}" "${DATASET_ID}"
            else
                remaining_days=$(( (SEVEN_DAYS - DIFF) / 86400 ))
                echo "距离下次历史合并还有约 ${remaining_days} 天"
            fi
        fi
        
        # 计算下次同步时间
        end_time=$(date +%s)
        duration=$((end_time - start_time))
        next_sync=$((SYNC_INTERVAL - duration))
        
        if [ $next_sync -gt 0 ]; then
            echo "同步完成，耗时 ${duration} 秒，下次同步将在 ${next_sync} 秒后 ($(date -d "@$(($(date +%s) + next_sync))" '+%Y-%m-%d %H:%M:%S'))"
            sleep $next_sync
        else
            echo "同步完成，耗时 ${duration} 秒 (超过同步间隔)，立即开始下次同步"
        fi
    done
}

# 启动OpenList服务
./openlist server &
SERVER_PID=$!

# 启动同步进程
if [ "$DISABLE_SYNC" != "true" ]; then
    sync_data &
    SYNC_PID=$!
fi

# 等待进程结束
wait $SERVER_PID

# 清理
if [ -n "$SYNC_PID" ]; then
    kill $SYNC_PID
fi

exit 0